David Lewin

Background

I am a Ph.D. candidate in statistics at the Wharton School of the University of Pennsylvania. I was recently working as the Director of Statistics for the Philadelphia Eagles. After trying to coordinate a full-time job and my dissertation, while still enjoying life with my one year old daughter, I decided to come back to Penn full-time to complete my degree. Before entering graduate school I was a math and computer science teacher at The Hun School of Princeton, as well as a coach for football and basketball. I had also worked at Johnson & Johnson as a supply chain analyst after I graduated from Cornell University with a degree in Operations Research & Industrial Engineering. Throughout the years I have also enjoyed working as tutor for all levels of math and statistics, from junior high to MBAs, as well as consulting for a variety of corporations in a range of industries. I enjoy cooking, wine, traveling, scuba diving, camping, skiing, and relaxing at home with my wife, daughter, and dog.

Research

My advisor at Penn is Professor Dylan Small. We have been working on causal inference and instrumental variable methods that apply to clinical trials when non-compliance among patients exists. Rather than excluding data, ignoring the fact that patients did not follow instructions, or analyzing the data based on which treatment the patients decided to receive, we are working on ways to improve the methods and thus the results in settings like this. In particular, we are focused on measuring outcomes related to patient safety, since this has been much discussed for drugs like Vioxx, and more research is needed to reduce these risks.

I have also been able to work with Dr. Sue Marcus of the Department of Psychiatry at the Mount Sinai School of Medicine. We collaborated with physicians to improve experimental designs for NIH funded studies, as well as researching causal inference methods. With Dr. Elaine Zanutto we examined missing data and methods for imputation for skewed, multivariate data. While an intern of Johnson & Johnson Pharmaceutical Research & Development, L.L.C. I worked with a variety of interesting biostatisticians and epidemiologists to examine data from a clinical trial as well as researching new methodology for multiple testing and comparing these to standard techniques.

Papers and Presentations

Lewin D and Small D. Using Instrumental Variables to Improve Safety Estimates from Clinical Trials. Presented at GlaxoSmithKline in Collegeville, PA; Johnson & Johnson in Titusville, NJ; Mid-Atlantic Causal Modeling Conference in Philadelphia, PA; and at the Conference of the International Society for Clinical Biostatistics in Geneva, Switzerland, 2006. Paper in progress.

Marcus S, Lewin D, Martinez J, Ray S, Mosovich S, Gorman L, Barlow D, Shear M, Woods S, Gorman J. A Comparison of Medication Side Effect Reports by Panic Disorder Patients With and Without Concomitant Cognitive Behavioral Therapy. The American Journal of Psychiatry 2007; 164: 273-275.

Marcus S, Tu X, Lewin D, Gibbons R. An Instrumental Variable Approach for Bounding Rater Bias. 2005 Proceedings for the Joint Statistical Meetings, Paper 353.

Marcus S, Zanutto E, Lewin D, Weaver J, Bannon W, McKay M, Irwin T, Morgenstern J. A Propensity Score Approach for Assessing Selection Bias in Uncontrolled Intervention Trials. Submitted.

Future Plans

I have had a wide range of experiences, and my goal upon receiving my doctorate is to combine my interests in statistical consulting, research, and educating. I would like to start my own statistical consulting firm, but I will also be considering positions in academia and at research organizations.